Disclosure of Invention
The embodiment of the invention provides a transformer load prediction method, a device thereof and electronic equipment, which at least solve the technical problem of lower accuracy of a prediction result in a mode of predicting the transformer load by configuring a weight value through an exponential smoothing method in the related technology.
According to one aspect of the embodiment of the invention, a transformer load prediction method is provided, and the method comprises the steps of collecting load data of a transformer in a target time period, preprocessing the load data to obtain preprocessed load data, calculating trend coefficients of each time point in the target time period based on the preprocessed load data, predicting the trend coefficients to obtain predicted trend coefficients of each time point in a future time period, calculating seasonal coefficients of the transformer, and predicting predicted load data of the transformer in the future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients.
The method comprises the steps of selecting a time point as a target observation point, obtaining a load value of the target observation point, obtaining a predicted load value of a historical observation point corresponding to the target observation point, configuring a smooth coefficient for the load value of the target observation point, calculating a trend coefficient of the target observation point based on the load value of the target observation point, the predicted load value of the historical observation point and the smooth coefficient, and repeating the steps until the trend coefficient of each time point in the target time period is calculated.
The method comprises the steps of firstly, constructing a linear regression model based on the trend coefficient and a time point corresponding to the trend coefficient, secondly, calculating a first regression coefficient and a second regression coefficient of the linear regression model, thirdly, calculating a predicted trend coefficient of a target time point based on the first regression coefficient, the second regression coefficient and the target time point to be predicted, and repeating the second to third until the sum of squares of residual errors of the linear regression model reaches a minimum value, so as to obtain the predicted trend coefficient of each time point in the future time period.
Optionally, the first regression coefficient expression of the linear regression model is: Where β1 denotes a first regression coefficient, n denotes the number of load data within the target period, Xi denotes a time point, Yi denotes a trend coefficient of the time point,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the second regression coefficient expression of the linear regression model is: Where beta0 represents the second regression coefficient, beta1 represents the first regression coefficient,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the step of calculating the seasonal factor of the transformer comprises the steps of obtaining an average load value of the transformer in a history period, dividing the history period by taking seasons as dividing targets to obtain the average load value of the transformer in each season, and calculating the ratio between the average load value of the transformer in each season and the average load value of the history period to obtain the seasonal factor of the transformer.
Optionally, the step of predicting the predicted load data of the transformer in the future time period based on the preprocessed load data, the predicted trend coefficient and the seasonal coefficient comprises the steps of calculating initial predicted load data of each time point based on the predicted trend coefficient and the load data, splitting the initial predicted load data into seasonal load data and non-seasonal load data, adjusting the seasonal load data based on the seasonal coefficient to obtain adjusted seasonal load data, and calculating final predicted load data based on the adjusted seasonal load data and the non-seasonal load data.
Optionally, the step of preprocessing the load data comprises filling missing values of the load data based on interpolation, detecting abnormal values of the load data, and eliminating the abnormal values.
According to another aspect of the embodiment of the invention, a transformer load prediction device is provided, which comprises an acquisition unit, a calculation unit, a first prediction unit and a second prediction unit, wherein the acquisition unit is used for acquiring load data of a transformer in a target time period and preprocessing the load data to obtain preprocessed load data, the calculation unit is used for calculating trend coefficients of each time point in the target time period based on the preprocessed load data, the first prediction unit is used for predicting the trend coefficients to obtain predicted trend coefficients of each time point in a future time period, and the second prediction unit is used for calculating seasonal coefficients of the transformer and predicting predicted load data of the transformer in the future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients.
The calculation unit comprises a first selection module, a first acquisition module, a first configuration module, a first calculation module and a first repetition module, wherein the first selection module is used for selecting a time point as a target observation point to acquire a load value of the target observation point, the first acquisition module is used for acquiring a predicted load value of a historical observation point corresponding to the target observation point, the first configuration module is used for configuring a smooth coefficient for the load value of the target observation point, the first calculation module is used for calculating a trend coefficient of the target observation point based on the load value of the target observation point, the predicted load value of the historical observation point and the smooth coefficient, and the first repetition module is used for repeating the first to the fourth steps until the trend coefficient of each time point in the target time period is calculated.
The first prediction unit comprises a first construction module, a second calculation module, a third calculation module and a second repetition module, wherein the first construction module is used for constructing a linear regression model based on the trend coefficient and a time point corresponding to the trend coefficient, the second calculation module is used for calculating a first regression coefficient and a second regression coefficient of the linear regression model, the third calculation module is used for calculating a predicted trend coefficient of the target time point based on the first regression coefficient, the second regression coefficient and a target time point to be predicted, and the second repetition module is used for repeating the second to the third steps until the sum of residual squares of the linear regression model reaches a minimum value, and the predicted trend coefficient of each time point in a future time period is obtained.
Optionally, the first regression coefficient expression of the linear regression model is: Where β1 denotes a first regression coefficient, n denotes the number of load data within the target period, Xi denotes a time point, Yi denotes a trend coefficient of the time point,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the second regression coefficient expression of the linear regression model is: Where beta0 represents the second regression coefficient, beta1 represents the first regression coefficient,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the second prediction unit comprises a second acquisition module, a first division module and a fourth calculation module, wherein the second acquisition module is used for acquiring an average load value of the transformer in a history period, the first division module is used for dividing the history period by taking seasons as division targets to obtain the average load value of the transformer in each season, and the fourth calculation module is used for calculating the ratio between the average load value of the transformer in each season and the average load value in the history period to obtain the seasonal coefficient of the transformer.
Optionally, the second prediction unit further comprises a fifth calculation module, a first splitting module, a first adjusting module and a sixth calculation module, wherein the fifth calculation module is used for calculating initial prediction load data of each time point based on the prediction trend coefficient and the load data, the first splitting module is used for splitting the initial prediction load data into seasonal load data and non-seasonal load data, the first adjusting module is used for adjusting the seasonal load data based on the seasonal coefficient to obtain adjusted seasonal load data, and the sixth calculation module is used for calculating final prediction load data based on the adjusted seasonal load data and the non-seasonal load data.
Optionally, the acquisition unit further comprises a first filling module for filling the missing value of the load data based on an interpolation method, and a first rejecting module for detecting the abnormal value of the load data and rejecting the abnormal value.
According to another aspect of the embodiments of the present invention, there is also provided an electronic device including one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement any one of the transformer load prediction methods described above.
Firstly, collecting load data of a transformer in a target time period, preprocessing the load data to obtain preprocessed load data, then calculating trend coefficients of each time point in the target time period based on the preprocessed load data, predicting the trend coefficients to obtain predicted trend coefficients of each time point in a future time period, finally calculating seasonal coefficients of the transformer, and predicting the predicted load data of the transformer in the future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients.
According to the application, for the load data in the historical time period, the trend coefficient of each time point is calculated, the trend coefficient is predicted, the seasonal variation is taken into consideration, the seasonal coefficient is calculated through the seasonal load data, the transformer load in the future time period is comprehensively predicted according to the seasonal coefficient, the original load data and the predicted trend coefficient, the trend coefficient can well reflect the change of the transformer in the long time period, the accuracy of the load prediction can be improved by combining the seasonal coefficient for comprehensive prediction, and the technical problem that in the related art, the transformer load is predicted by configuring the weight value through an exponential smoothing method, and the accuracy of the prediction result is lower is solved.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be noted that the method and the device for predicting the load of the transformer in the application can be used in the technical field of power grid safety under the condition of predicting the load value of the transformer, and can also be used in any field except the technical field of power grid safety under the condition of predicting the load value of the transformer.
The following embodiments of the present invention are applicable to various transformer load prediction systems/applications/devices. The method utilizes the trend and seasonal information in the historical data to predict the load of the transformer, has higher prediction precision and stability, can more accurately capture the long-term trend and seasonal change in the load data, thereby improving the prediction accuracy, can better adapt to the change and uncertainty of the load data, flexibly adapt to different types of load data, is not limited by the data distribution and characteristics, can process the data with different scales, different frequencies and different time periods, and can be suitable for complex change modes of the load of the transformer.
The present invention will be described in detail with reference to the following examples.
Example 1
According to an embodiment of the present invention, there is provided an embodiment of a transformer load prediction method, it being noted that the steps shown in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order different from that shown or described herein.
Fig. 1 is a flowchart of an alternative transformer load prediction method according to an embodiment of the present invention, as shown in fig. 1, the method comprising the steps of:
step S101, collecting load data of a transformer in a target time period, and preprocessing the load data to obtain preprocessed load data;
step S102, calculating trend coefficients of each time point in the target time period based on the preprocessed load data;
Step S103, predicting trend coefficients to obtain predicted trend coefficients of each time point in a future time period;
step S104, calculating seasonal factors of the transformer, and predicting predicted load data of the transformer in a future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal factors.
Through the steps, firstly, load data of the transformer in a target time period are collected, the load data are preprocessed to obtain preprocessed load data, then, trend coefficients of each time point in the target time period are calculated based on the preprocessed load data, then, the trend coefficients are predicted to obtain predicted trend coefficients of each time point in a future time period, finally, seasonal coefficients of the transformer are calculated, and predicted load data of the transformer in the future time period are predicted based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients.
In this embodiment, for the load data in the historical time period, the trend coefficient of each time point is calculated, the trend coefficient is predicted, the seasonal variation is considered, the seasonal coefficient is calculated through the seasonal load data, the transformer load in the future time period is comprehensively predicted according to the seasonal coefficient, the original load data and the predicted trend coefficient, the trend coefficient can well reflect the change of the transformer in the long time period, the comprehensive prediction is combined with the seasonal coefficient, the accuracy of the load prediction can be improved, and further the technical problem that in the related technology, the transformer load is predicted by configuring the weight value through an exponential smoothing method, and the accuracy of the prediction result is low is solved.
Embodiments of the present invention will be described in detail with reference to the following steps.
It should be noted that load data of a transformer may refer to current and power loads to which the transformer is subjected during operation, and these data are critical for evaluating the performance of the transformer. The load data may include active power and reactive power for reflecting the load change law of the transformer in different time periods, helping to obtain the operating state and efficiency of the transformer, and may also be used to evaluate the characteristics of the transformer, such as the resistive and inductive parameters obtained through open circuit tests and short circuit tests, and may evaluate the efficiency, temperature rise, life, and parallel operation capability of the transformer. In a word, the load data of the transformer is an important means for evaluating the characteristics of the transformer, and through analysis and application of the load data, the operation efficiency and reliability of the transformer can be improved, the stable operation of the power system is ensured, and the stability of the power system is improved.
In the planning, operation and scheduling processes of the power system, the transformer load is predicted, so that the state and characteristic change of the transformer can be predicted in advance, and further, power staff can be assisted in reasonably configuring resources, thereby improving the power grid operation efficiency and reducing the operation cost.
Step S101, collecting load data of the transformer in a target time period, and preprocessing the load data to obtain preprocessed load data.
It should be noted that, as an electrical device, a transformer is used to raise or lower the values of ac voltage and current based on the principle of electromagnetic induction, while transmitting electric energy, and is an indispensable component for power transmission and distribution. The load data can comprise load current, active power, reactive power, apparent power and load rate, wherein the load current is the value of each phase line current of the transformer in the operation process, the active power refers to electric energy actually consumed by the transformer, the reactive power is related to the active power but does not actually consume a power part of the electric energy, the apparent power refers to the total power capacity of the transformer, the active power and the reactive power are combined, the load rate refers to the ratio of the actual load to the rated capacity of the transformer, and the load data can be used for evaluating the operation efficiency of the transformer.
After the load data of the transformer are collected, the load data are preprocessed, and the missing value and the abnormal value in the load data are processed, so that the integrity and the accuracy of the load data are ensured.
Optionally, the step of preprocessing the load data comprises filling the missing value of the load data based on an interpolation method, detecting the abnormal value of the load data, and eliminating the abnormal value.
It should be noted that, in the embodiment of the present invention, the missing value of the load data may be filled by an interpolation method to ensure the continuity and integrity of the data, where the interpolation formula is expressed as follows: Wherein Yt is a deletion value, and Yt-1 and Yt+1 are the most recent observations before and after the deletion value, respectively.
It should be noted that, in the embodiment of the present invention, the abnormal value may be detected by using a box diagram to select and reject or correct the abnormal value in the load data, where the box diagram formula is expressed as:
IQR=Q3-Q1
Lower_Bound=Q1-1.5×IQR
Upper_Bound=Q3-1.5×IQR
Where IQR is the quartile spacing, Q1 is the first quartile, Q3 is the third quartile, lower_bound is the Lower limit, upper_bound is the Upper limit, and outliers are defined as observations below or above the Lower limit.
Step S102, calculating trend coefficients of each time point in the target time period based on the preprocessed load data.
It should be noted that, in the embodiment of the present invention, the trend coefficient of each time point in the target time period is calculated to predict the load value of the transformer, and the trend coefficient may reflect the load change condition of the transformer.
The method comprises the steps of selecting a time point as a target observation point, obtaining a predicted load value of a historical observation point corresponding to the target observation point, configuring a smooth coefficient for the load value of the target observation point, calculating a trend coefficient of the target observation point based on the load value of the target observation point, the predicted load value of the historical observation point and the smooth coefficient, and repeating the steps from the first step to the fourth step until the trend coefficient of each time point in the target time period is calculated.
Specifically, when calculating the trend coefficient, firstly, selecting a time point as a target observation point, then obtaining a load value of the target observation point, simultaneously obtaining a predicted load value of the last time point corresponding to the target observation point, comprehensively calculating the trend coefficient of the target observation point according to a preset smooth coefficient (which can be a numerical value between 0 and 1), quantifying the change trend of the transformer at the target observation point by using the trend coefficient, and repeating the steps until the trend coefficient of all time points in the target time period is calculated, thereby obtaining the trend coefficient of each time point in the target time period.
Specifically, the calculation formula of the trend coefficient is expressed as Tt＝α·Yt+(1-α)·Tt-1. Wherein Tt is a trend coefficient of the transformer at the target observation point T, Yt is load data of the transformer at the target observation point, α is a smoothing coefficient, and generally takes a value between 0 and 1, and represents a weight on a past observation value, and Tt-1 is a trend coefficient of a previous observation point of the target observation point.
Step S103, predicting the trend coefficient to obtain a predicted trend coefficient of each time point in the future time period.
It should be noted that, the embodiment of the invention predicts the trend coefficient, so as to obtain the predicted trend coefficient of each time point in the target time period, and then comprehensively calculates according to the load data of each time point in the target time period to obtain the predicted load value of each time point in a future time period, thereby realizing the prediction of the load value of the transformer, the predicted trend coefficient is not limited by the data distribution and the characteristics, can flexibly adapt to different types of load data, and improves the application range of the prediction and the accuracy of the prediction result.
The method comprises the steps of firstly, constructing a linear regression model based on the trend coefficient and a time point corresponding to the trend coefficient, secondly, calculating a first regression coefficient and a second regression coefficient of the linear regression model, thirdly, calculating a predicted trend coefficient of a target time point based on the first regression coefficient and the second regression coefficient and the target time point to be predicted, and repeating the steps of secondly to thirdly until the sum of squares of residuals of the linear regression model reaches a minimum value, so as to obtain the predicted trend coefficient of each time point in the future time period.
When predicting the trend coefficient of each time point, firstly taking the trend coefficient as a dependent variable, taking the time point corresponding to the trend coefficient as an independent variable, constructing a linear regression model, wherein the linear regression model expression comprises the independent variable, the dependent variable, the first regression coefficient and the second regression coefficient, then calculating the values of the first regression coefficient and the second regression coefficient according to a calculation formula of the regression coefficients, substituting the values of the first regression coefficient and the second regression coefficient into the linear regression model, calculating the predicted trend coefficient corresponding to each time point, repeating the steps for iterative calculation until the residual square sum of the model is minimum, and obtaining the predicted trend coefficient of each time point finally.
Specifically, the embodiment of the invention can take the trend coefficient as a dependent variable and the time point as an independent variable, and predict the trend coefficient of each time point by fitting a linear regression model, wherein the formula of the linear regression model is as follows:
Y=β0+β1X+∈
Where Y is the dependent variable (trend coefficient), X is the independent variable (time point), β0 (first regression coefficient) and β1 (second regression coefficient) are regression coefficients, ε is the error term;
optionally, the first regression coefficient expression of the linear regression model isWhere β1 denotes a first regression coefficient, n denotes the number of load data within the target period, Xi denotes a time point, Yi denotes a trend coefficient of the time point,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the second regression coefficient expression of the linear regression model is: Where beta0 represents the second regression coefficient, beta1 represents the first regression coefficient,The average value of the time points is indicated,Represents the average value of the trend coefficients.
When predicting trend coefficients of each time point based on the linear regression model, calculating values of regression coefficients beta0 and beta1 by using the following formulas, so that the sum of squares of residual errors of the model is minimized, obtaining a predicted trend coefficient of each time point finally, wherein the calculation formulas of the second regression coefficient and the first regression coefficient are respectively expressed as follows:
Wherein, theAndThe independent variable and the average value of the dependent variable, respectively, and n is the number of samples.
Step S104, calculating seasonal factors of the transformer, and predicting predicted load data of the transformer in a future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal factors.
After the trend coefficient of each time point is predicted, the embodiment of the invention takes the seasonal factor as one of the indexes of the load value prediction, captures the seasonal factor in the historical load data by calculating the seasonal factor, can improve the accuracy of the calculation result, and comprehensively calculates the predictive load data in the future time period by taking the nodular factor, the load data of each time point and the predicted trend coefficient as the load value prediction indexes.
Optionally, the step of calculating the seasonal factor of the transformer comprises the steps of obtaining an average load value of the transformer in a history period, dividing the history period by taking seasons as dividing targets to obtain the average load value of the transformer in each season, and calculating the ratio between the average load value of the transformer in each season and the average load value of the history period to obtain the seasonal factor of the transformer.
When calculating the seasonal factor, firstly, calculating an average load value of each season in an observation period (corresponding to the history period), dividing the observation period into a plurality of seasons by a unit of a season, calculating an average load value of each season, and calculating a ratio between the average load value of each season and the average load value of the history period to obtain the seasonal factor corresponding to each season.
The method comprises the steps of calculating initial predicted load data of each time point based on pre-processed load data, predicted trend coefficients and seasonal coefficients, splitting the initial predicted load data into seasonal load data and non-seasonal load data, adjusting the seasonal load data based on the seasonal coefficients to obtain adjusted seasonal load data, and calculating final predicted load data based on the adjusted seasonal load data and non-seasonal load data.
When the load value of the transformer in a short time in the future is comprehensively predicted based on the preprocessed load data, the predicted trend coefficient and the seasonal coefficient, the predicted trend coefficient and the load data at each time point are accumulated to obtain initial predicted load data at each time point, then the initial predicted load data are adjusted according to the seasonal coefficient, specifically, the calculated initial predicted overload data are split into seasonal load data and non-seasonal load data, the seasonal coefficient corresponding to the target time period is obtained, the product of the seasonal coefficient and the seasonal load data is calculated to obtain adjusted seasonal load data, and finally the adjusted seasonal coincidence data and the non-seasonal load data are accumulated to obtain final predicted load data, so that the prediction of the load data of the transformer is completed.
The following detailed description is directed to alternative embodiments.
FIG. 2 is a schematic diagram of an alternative transformer load prediction process according to an embodiment of the invention, as shown in FIG. 2, the transformer load prediction process comprising:
Step one, collecting historical load data of a transformer, including load values of each hour or each day, carrying out missing value processing and abnormal value detection on the historical load data, filling the missing values and eliminating the abnormal values;
Historical load data of the transformer, including hourly or daily load amounts, is collected and may be obtained by a transformer monitoring system, sensors or historian.
The missing values are filled in using interpolation methods to ensure data continuity and integrity. In this embodiment, a linear interpolation method is selected and used, and the linear interpolation formula is as follows: Wherein Yt is a deletion value, and Yt-1 and Yt+1 are the most recent observations before and after the deletion value, respectively.
Detecting outliers using statistical methods (e.g., box plot or Zscore) or machine learning methods (e.g., isolated forest or cluster-based methods), and for detected outliers, culling or correction may be selected, in this embodiment, the outliers are culled in a box plot, which is illustrated by the formula:
IQR=Q3-Q1
Lower_Bound=Q1-1.5×IQR
Upper_Bound=Q3-1.5×IQR
Where IQR is the quartile spacing, Q1 is the first quartile, Q3 is the third quartile, lower_bound is the Lower limit, upper_bound is the Upper limit, and outliers are defined as observations below or above the Lower limit.
The missing values are filled in through the steps to ensure the continuity and the integrity of the data, and then abnormal values are detected and removed to reduce the interference to the establishment of a subsequent model.
Estimating the trend of the original load data, and calculating the trend coefficient of each time point;
in calculating the trend coefficient at each time point, an initial value is first selected as a starting point (corresponding to the above-described target observation point), then a new estimated value is calculated from the load data of the starting point and the estimated value at the last time point, and the trend estimated value at each time point is calculated using the following formula:
Tt＝α·Yt+(1-α)·Tt-1
Wherein Tt is a trend estimated value at time T, Yt is an original load value at time T, α is a smoothing coefficient, and generally takes a value between 0 and 1, which represents a weight on a past observed value, and the obtained trend estimated value is a trend coefficient, which represents a trend at each time point.
Predicting the trend coefficient, and identifying a long-term trend to obtain a predicted trend coefficient of each time point;
The embodiment of the invention predicts the trend coefficient by using a linear regression model to identify the long-term trend. In this embodiment, the trend coefficient is regarded as a dependent variable, the time is regarded as an independent variable, and the future trend coefficient is predicted by fitting a linear regression model, which is formulated as:
Y=β0+β1X+∈
Where Y is the dependent variable (trend coefficient), X is the independent variable (time), β0 and β1 are the first regression coefficient and the second regression coefficient, respectively, and ε is the error term.
The values of the regression coefficients β0 and β1 are estimated using the following formula, so that the sum of squares of the residuals of the model is minimized, and the final predicted trend coefficients at each time point are obtained by continuous fitting:
Wherein, theAndThe independent variable and the average value of the dependent variable, respectively, and n is the number of samples.
Step four, adding the predicted trend coefficient into the original load data to obtain adjusted load data;
And (3) corresponding the predicted trend coefficient predicted in the step (III) to a time point, and adding the predicted trend coefficient to the original load data to obtain adjusted load data, wherein a calculation formula is expressed as follows:
Wherein, theIs the adjusted load data, i.e. the initial load predicted value of the transformer at time T, Yt is the original load value at time T, and Tt is the predicted trend coefficient at time T.
Through the steps, the predicted trend is added to the original load data to obtain more accurate adjusted load data. This process is logically associated with the trend prediction step, providing the basis for the final load prediction.
Step five, carrying out seasonal decomposition on the adjusted load data to capture seasonal variation and obtain seasonal data (corresponding to the seasonal load data);
In this embodiment, the adjusted load data is decomposed into trends, seasonality and residuals to capture seasonal variations, and the seasonal decomposition formula is expressed as:
Wherein, theIs the adjusted initial predicted load data at time T, Tt is the trend portion, St is the seasonal portion, and Rt is the residual portion.
Specifically, the trend portion Tt is estimated by a moving average or other smoothing method, then the remainder, Y't-Tt, is obtained by calculating the difference between the raw load data and the trend portion, and finally the remainder is seasonally analyzed to obtain a seasonally portion St (corresponding to the seasonally data).
Through this step, the initial predicted load data of the transformer may be seasonally decomposed using an additive model to capture seasonal variations.
And step six, calculating seasonal coefficients, and adjusting the seasonal data based on the seasonal coefficients to obtain final predicted load data.
Selecting an observation period, for example, a whole year can be selected as an observation period, calculating an average load value of a transformer in the observation period, dividing the observation period according to seasons to obtain an average load value of each season, and calculating a ratio of the average load value of each season to the average load value of the observation period to obtain a seasonal factor of each season.
After the seasonal factor is obtained, the seasonal factor is used for adjusting the seasonal data so as to ensure the accuracy of the seasonal variation, and the expression of the seasonal adjustment is as follows:
Where S't is the adjusted seasonal data, St is the original seasonal data at time t, It is the seasonal factor at time t,Is the average of all seasonal indices.
And finally, accumulating the adjusted seasonal data with the trend part and the residual part which are obtained by splitting, and calculating to obtain final predicted load data.
The embodiment of the invention predicts the load of the transformer by utilizing the trend and seasonal information in the historical data, has higher prediction precision and stability, can more accurately capture the long-term trend and seasonal change in the load data, thereby improving the prediction accuracy, and the data driving method can better adapt to the change and uncertainty of the load data, flexibly adapt to different types of load data, is not limited by data distribution and characteristics, can process the data with different scales, different frequencies and different time periods, and can be suitable for complex change modes of the load of the transformer.
The following describes in detail another embodiment.
Example two
The transformer load prediction device provided in this embodiment includes a plurality of implementation units, where each implementation unit corresponds to each implementation step in the first embodiment, and specific implementation and beneficial effects of each implementation unit may refer to the foregoing method embodiment and will not be described herein.
Fig. 3 is a schematic view of an alternative transformer load predicting apparatus according to an embodiment of the present invention, which may include, as shown in fig. 3, an acquisition unit 31, a calculation unit 32, a first prediction unit 33, a second prediction unit 34, wherein,
The collecting unit 31 is configured to collect load data of the transformer in a target time period, and perform preprocessing on the load data to obtain preprocessed load data;
a calculation unit 32 for calculating a trend coefficient for each time point in the target period based on the preprocessed load data;
a first prediction unit 33, configured to predict the trend coefficient to obtain a predicted trend coefficient for each time point in the future time period;
And a second prediction unit 34 for calculating a seasonal coefficient of the transformer and predicting predicted load data of the transformer in a future period based on the preprocessed load data, the predicted trend coefficient, and the seasonal coefficient.
The transformer load prediction device is used for acquiring load data of a transformer in a target time period through the acquisition unit 31 and preprocessing the load data to obtain preprocessed load data, calculating trend coefficients of each time point in the target time period through the calculation unit 32 based on the preprocessed load data, predicting the trend coefficients through the first prediction unit 33 to obtain predicted trend coefficients of each time point in a future time period, calculating seasonal coefficients of the transformer through the second prediction unit 34, and predicting the predicted load data of the transformer in the future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients.
In this embodiment, for the load data in the historical time period, the trend coefficient of each time point is calculated, the trend coefficient is predicted, the seasonal variation is considered, the seasonal coefficient is calculated through the seasonal load data, the transformer load in the future time period is comprehensively predicted according to the seasonal coefficient, the original load data and the predicted trend coefficient, the trend coefficient can well reflect the change of the transformer in the long time period, the comprehensive prediction is combined with the seasonal coefficient, the accuracy of the load prediction can be improved, and further the technical problem that in the related technology, the transformer load is predicted by configuring the weight value through an exponential smoothing method, and the accuracy of the prediction result is low is solved.
Optionally, the calculating unit 32 includes a first selecting module configured to select a time point as a target observation point to obtain a load value of the target observation point, a first obtaining module configured to obtain a predicted load value of a historical observation point corresponding to the target observation point, a first configuring module configured to configure a smoothing coefficient for the load value of the target observation point, a first calculating module configured to calculate a trend coefficient of the target observation point based on the load value of the target observation point, the predicted load value of the historical observation point, and the smoothing coefficient, and a first repeating module configured to repeat the steps one to four until the trend coefficient of each time point in the target time period is calculated.
Optionally, the first prediction unit 33 includes a first construction module configured to construct a linear regression model based on the trend coefficient and the time point corresponding to the trend coefficient, a second calculation module configured to calculate a first regression coefficient and a second regression coefficient of the linear regression model, a third calculation module configured to calculate a predicted trend coefficient of the target time point based on the first regression coefficient and the second regression coefficient and the target time point to be predicted, and a second repetition module configured to repeat the steps two to three until the sum of squares of residuals of the linear regression model reaches a minimum value, and obtain the predicted trend coefficient of each time point in the future time period.
Optionally, the first regression coefficient expression of the linear regression model is: Where β1 denotes a first regression coefficient, n denotes the number of load data within the target period, Xi denotes a time point, Yi denotes a trend coefficient of the time point,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the second regression coefficient expression of the linear regression model is: Where beta0 represents the second regression coefficient, beta1 represents the first regression coefficient,The average value of the time points is indicated,Represents the average value of the trend coefficients.
Optionally, the second prediction unit 34 includes a second obtaining module, configured to obtain an average load value of the transformer in a history period, a first dividing module, configured to divide the history period with seasons as a division target to obtain average load values of the transformer in each season, and a fourth calculating module, configured to calculate a ratio between the average load values of the transformer in each season and the average load values of the history period to obtain seasonal factors of the transformer.
Optionally, the second prediction unit 34 further includes a fifth calculation module for calculating initial predicted load data at each time point based on the predicted trend coefficient and the load data, a first splitting module for splitting the initial predicted load data into seasonal load data and non-seasonal load data, a first adjustment module for adjusting the seasonal load data based on the seasonal coefficient to obtain adjusted seasonal load data, and a sixth calculation module for calculating final predicted load data based on the adjusted seasonal load data and non-seasonal load data.
Optionally, the acquisition unit 31 further includes a first filling module for filling the missing value of the load data based on interpolation, and a first rejecting module for detecting the abnormal value of the load data and rejecting the abnormal value.
The transformer load predicting device may further include a processor and a memory, wherein the acquisition unit 31, the calculation unit 32, the first prediction unit 33, the second prediction unit 34, and the like are stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions.
The processor includes a kernel, and the kernel fetches a corresponding program unit from the memory. The kernel may be provided with one or more kernel parameters that are adjusted to predict the load value of the transformer.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), which includes at least one memory chip.
According to another aspect of the embodiment of the present invention, there is also provided a computer readable storage medium, including a stored computer program, where the computer program is configured to control a device in which the computer readable storage medium is located to perform any one of the above-mentioned transformer load prediction methods when the computer program is executed.
According to another aspect of the embodiments of the present invention, there is also provided an electronic device including one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement any one of the transformer load prediction methods described above.
According to another aspect of the embodiments of the present invention, there is also provided a computer program product comprising a computer program, wherein the computer program when executed by a processor implements any one of the transformer load prediction methods described above.
The application also provides a computer program product adapted to perform, when executed on a data processing apparatus, a program initialized with the method steps of collecting load data of a transformer in a target time period and preprocessing the load data to obtain preprocessed load data, calculating trend coefficients for each time point in the target time period based on the preprocessed load data, predicting trend coefficients to obtain predicted trend coefficients for each time point in a future time period, calculating seasonal coefficients of the transformer, and predicting predicted load data of the transformer in the future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients.
The method comprises the steps of selecting a time point as a target observation point, obtaining a predicted load value of a historical observation point corresponding to the target observation point, configuring a smooth coefficient for the load value of the target observation point, calculating a trend coefficient of the target observation point based on the load value of the target observation point, the predicted load value of the historical observation point and the smooth coefficient, and repeating the steps from the first step to the fourth step until the trend coefficient of each time point in the target time period is calculated.
The application further provides a computer program product which is suitable for executing a program initialized with the following method steps when the program is executed on data processing equipment, wherein the step of predicting trend coefficients to obtain predicted trend coefficients of each time point in a future time period comprises the steps of firstly, constructing a linear regression model based on the trend coefficients and the time points corresponding to the trend coefficients, secondly, calculating a first regression coefficient and a second regression coefficient of the linear regression model, thirdly, calculating predicted trend coefficients of target time points based on the first regression coefficient and the second regression coefficient and target time points to be predicted, and repeating the steps from the second to the third until the sum of residual squares of the linear regression model reaches a minimum value to obtain the predicted trend coefficients of each time point in the future time period.
The application also provides a computer program product adapted to perform, when executed on a data processing apparatus, a program initialized with the method steps of: Where β1 denotes a first regression coefficient, n denotes the number of load data within the target period, Xi denotes a time point, Yi denotes a trend coefficient of the time point,The average value of the time points is indicated,Represents the average value of the trend coefficients.
The application also provides a computer program product adapted to perform, when executed on a data processing apparatus, a program initialized with the method steps of: Where beta0 represents the second regression coefficient, beta1 represents the first regression coefficient,The average value of the time points is indicated,Represents the average value of the trend coefficients.
The application also provides a computer program product adapted to perform, when executed on a data processing device, a program initialized with the method steps of obtaining a seasonal factor of a transformer, the step of obtaining an average load value of the transformer over a history period, dividing the history period with a season as a division target to obtain an average load value of the transformer over each season, and calculating a ratio between the average load value of the transformer over each season and the average load value over the history period to obtain the seasonal factor of the transformer.
The application also provides a computer program product adapted to perform, when executed on a data processing apparatus, a program initialized with the method steps of predicting predicted load data of a transformer in a future time period based on the preprocessed load data, the predicted trend coefficients and the seasonal coefficients, comprising calculating initial predicted load data for each time point based on the predicted trend coefficients and the load data, splitting the initial predicted load data into seasonal load data and non-seasonal load data, adjusting the seasonal load data based on the seasonal coefficients to obtain adjusted seasonal load data, and calculating final predicted load data based on the adjusted seasonal load data and non-seasonal load data.
The application also provides a computer program product adapted to perform, when executed on a data processing apparatus, an initialization routine having the method steps of preprocessing load data comprising filling in missing values of the load data based on interpolation, detecting outliers of the load data, and rejecting the outliers.
Fig. 4 is a block diagram of a hardware structure of an electronic device (or a mobile device) of a transformer load prediction method according to an embodiment of the present invention. As shown in fig. 4, the electronic device may include one or more processors 402 (shown in fig. 4 as 402a, 402b,) 402n (the processor 402 may include, but is not limited to, a microprocessor MCU or a processing device such as a programmable logic device FPGA), a memory 404 for storing data. Among other things, a display, an input/output interface (I/O interface), a Universal Serial Bus (USB) port (which may be included as one of the ports of the I/O interface), a network interface, a keyboard, a power supply, and/or a camera may be included. It will be appreciated by those of ordinary skill in the art that the configuration shown in fig. 4 is merely illustrative and is not intended to limit the configuration of the electronic device described above. For example, the electronic device may also include more or fewer components than shown in FIG. 4, or have a different configuration than shown in FIG. 4.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes a U disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, etc. which can store the program code.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.